Method, apparatus, and computer-readable storage medium for optimizing slice resources for wireless communication.
Patent Information
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- ZTE CORP
- Filing Date
- 2023-07-14
- Publication Date
- 2026-08-03
Smart Images
Figure 2026525737000001_ABST
Abstract
Description
Technical Field
[0001] Technical Field The present disclosure generally relates to wireless communication, and more particularly to the optimization of RAN slice resources.
Background Art
[0002] Background Wireless communication technology is an increasingly central component of the interconnected global communication network. Wireless communication relies on time resources and frequency resources that are precisely allocated to transmit and receive wireless signals. Radio Access Network (RAN) resource slicing involves partitioning RAN resources into multiple slices or segments for different uses and services. For example, a slice can be allocated a specific set of resources for a particular service or user for which it provides the service. The allocation and optimization of the allocation of RAN slice resources is a problem for the efficient use of the communication resources of a communication system.
Summary of the Invention
Means for Solving the Problems
[0003] Summary This summary is a brief description of certain aspects of the present disclosure. This summary is not intended to limit the scope of the present disclosure.
[0004] According to some embodiments of this disclosure, a wireless communication method is disclosed. The method includes acquiring input information and using the input information to perform at least one of model training of a slice radio resource management model or generating an inference output using a slice radio resource management model. The input information includes at least one of slice measurement result information, one or more first pieces of information, one or more second pieces of information, or inference feedback information. The slice measurement result information includes at least one of UE location information, UE measurement information, or UE capability information. One or more first pieces of information include first current or predicted slice available capacity information, first current or predicted shared network slice information, first current or predicted preferred network slice information, first current or predicted dedicated network slice information, first multi-carrier resource sharing configuration information, first slice radio resource management (RRM) policy or restriction information, resource deployment information, slice-based cell reselection information, slice service level agreement information, PDU session quality of service (QoS) information, information on the attributes of slice resources used, predicted service traffic information, dual connectivity configuration information, validity time information, or reliability information. One or more of the second pieces of information include at least one of the following: second current or predicted slice available capacity information, second current or predicted shared network slice information, second current or predicted preferred network slice information, second current or predicted dedicated network slice information, second multi-carrier resource sharing configuration information, or second slice radio resource management (RRM) policy or restriction information.
[0005] A further embodiment of the present disclosure provides a wireless communication device comprising one or more memory units for storing one or more programs, and one or more processors electrically coupled to one or more memory units and configured to execute one or more programs and perform any method or step or combination thereof in the present disclosure.
[0006] A further embodiment of the present disclosure provides a non-temporary computer-readable storage medium for storing one or more programs, wherein the one or more programs are configured to cause at least one processor to perform any method or step or combination thereof as described in the present disclosure.
[0007] According to some embodiments of this disclosure, one or more wireless communication methods are further disclosed, the methods including combinations of specific methods, aspects, elements, and steps (in either a general or specific view) disclosed in various embodiments or examples of this disclosure.
[0008] The above and other embodiments, as well as their implementations, will be described in more detail in the drawings, this description, and the claims. [Brief explanation of the drawing]
[0009] Various exemplary embodiments of this disclosure are described in detail below with reference to the following drawings. The drawings are provided for illustrative purposes only and merely depict exemplary embodiments of this disclosure to facilitate understanding of this disclosure. Therefore, the drawings should not be considered to limit the breadth, scope, or applicability of this disclosure. It should be noted that these drawings are not necessarily drawn to scale in order to make the illustration clear and easy.
[0010] [Figure 1A] Figure 1A shows the UE handover from the source RAN node to the target RAN node. [Figure 1B] Figure 1B shows the UE handover from the source RAN node to the target RAN node.
[0011] [Figure 2] Figure 2 illustrates the functional framework of the RAN intelligent slice resource management system.
[0012] [Figure 3] Figure 3 shows a flowchart of wireless communication according to several embodiments of the present disclosure.
[0013] [Figure 4] Figure 4 shows another flowchart of wireless communication according to some embodiments of the present disclosure.
[0014] [Figure 5] Figure 5 shows another flowchart of wireless communication according to some embodiments of the present disclosure.
[0015] [Figure 6] Figure 6 shows another flowchart of wireless communication according to some embodiments of the present disclosure.
[0016] [Figure 7] Figure 7 shows the structure of the wireless communication system. [Modes for carrying out the invention]
[0017] Detailed explanation The rapid development of mobile communications has permeated people's work, social life, and every aspect of their lives, significantly impacting their lifestyles, work styles, socio-political and economic aspects, and other areas. Human society has entered the information age, and business application requirements in all areas are showing explosive growth. In the future, mobile networks may not only provide communication between people, but also services for mass devices and other purposes within the Internet of Things.
[0018] Differentiated business models and service requirements pose significant challenges for future wireless mobile broadband systems with respect to frequency, technology, and operation. The traditional communication network, namely, Residential Access Network (RAN) plus CORE, is becoming increasingly unable to meet all scenarios.
[0019] The development of Network Function Virtualization (NFV) technology enables operators to build different virtual networks for different business requirements. Network slicing is based on a common physical infrastructure to logically define and partition the network to form end-to-end virtual networks. Each virtual network has different functions and characteristics to dynamically meet various needs and business models.
[0020] A normal network slice includes a set of virtualized access network functions and core network functions. Network slicing can be constructed by operators according to requirements and strategies, and the functions included in a network slice are also determined by operators according to requirements and strategies. For example, some network slices may include a dedicated transfer plane in addition to control plane functions, and some network slices may only include some basic control plane functions.
[0021] As shown in FIGS. 1A and 1B, the ongoing (one or more) slices of the UE (User Equipment) can be supported by both the source NG-RAN node and the target NG-RAN node when the UE switches from one base station (BS1) to another base station (BS2). However, if the target RAN is overloaded such that it does not have sufficient slice resources for the incoming UE, the connection of the UE may be interrupted. To ensure the service continuity of the UE, achieving better slice resource allocation by the RAN nodes is a technical problem to be overcome. The AI (Artificial Intelligence) function can be a candidate approach for slice resource allocation prediction based on the collected data and for training / inference by the AI model.
[0022] FIG. 2 illustrates the functional framework of a RAN intelligent slice resource management system. The RAN intelligent slice resource management system includes a data collector, a model trainer, a model inference unit, and an actor. The data collector receives feedback information from the actor and provides input data to the model trainer and the model inference unit. The model trainer is configured to perform model training such as training, validating, and testing an artificial intelligence (AI) model or a machine learning (ML) model, which can generate model performance metrics as part of the model testing procedure. The training can include, for example, supervised learning, unsupervised learning, reinforcement learning, transfer learning, semi-supervised learning, or self-supervised learning. AI / ML inference can generally include a model or algorithm that represents knowledge or patterns learned from data, where the training performed by the trainer can be used to prepare the model. Inference data is supplied to the model to produce an output or prediction. The AI / ML model here can include, for example, at least one of an artificial neural network, a decision tree, a support vector machine, a gradient boosting model, an ensemble model, a generative model, and / or a probabilistic model.
[0023] The model inference unit is configured to provide AI / ML model inference output based on inference data provided by the data collector. The output can be predictive or deterministic. Inference may include, for example, online / real-time inference, batch inference, and edge inference. The output of the model inference unit may include one or more policies for actors such as core networks or base stations, which use one or more such policies to allocate slice resources in their communication systems. Furthermore, the actors are configured to monitor the performance of implementing the output of the model inference unit and feedback-related information to the data collector for future training.
[0024] According to some embodiments, the AI / ML model trainer can be located in the OAM (Operations, Management, and Maintenance) of the core network (CN), and the AI / ML model inference unit can be located in the gNB (or base station). Alternatively or additionally, both the AI / ML model trainer and the AI / ML model inference unit can be located in the gNB.
[0025] Alternatively or additionally, according to some embodiments of the gNB-CU (centralized unit) and gNB-DU (distributed unit) structures, the AI / ML model trainer can be located in the OAM and the AI / ML model inference unit can be located in the gNB-CU. Alternatively or additionally, both the AI / ML model trainer and the AI / ML model inference unit can be located in the gNB-CU. Alternatively or additionally, the AI / ML model trainer can be located in the gNB-CU and the AI / ML model inference unit can be located in the gNB-DU.
[0026] AI / ML-based RAN slice resource allocation input According to some embodiments, the input of information received by the AI / ML model trainer and / or the input of the AI / ML model inference unit may include the following:
[0027] The inputs (e.g., support information) provided to the AI / ML model trainer or AI / ML model inference unit (or data collector) from the local node where the AI / ML model trainer or AI / ML model inference unit is located include at least one or a combination of the following: first current or predicted slice availability information, first current or predicted shared network slice information, first current or predicted preferred network slice information, first current or predicted dedicated network slice information, first multi-carrier resource sharing configuration information, first slice radio resource management (RRM) policy or restriction information, resource deployment information, slice-based cell reselection information, slice service level agreement information, PDU session quality of service (QoS) information, information on attributes of slice resources used, predicted service traffic information, dual connectivity configuration information, validity time information, or confidence information. For example, current or predicted slice availability information indicates the capacity of available slice sources on the local network node. The information may reflect the current state or may be a predicted value (i.e., predicted or current value). Current or predicted shared network slice information can indicate the status of shared network slices, such as the usage of shared slices including PRBs, PRB ULs, PRB DLs, RRC connected users, and DRBs. The information can reflect the current status or represent predicted values. Current or predicted preferred network slice information can indicate the status of preferred network slices, such as the usage of shared slices including PRBs (Physical Resource Blocks), PRB ULs (Uplinks), PRB DLs (Downlinks), RRC (Radio Resource Control) connected users, and DRBs (Data Radio Bearers). The information can reflect the current status or represent predicted values. The information can reflect the current status or represent predicted values.Current or forecast-only network slice information can indicate the status of preferred network slices, such as the use of shared slices including PRB, PRB UL, PRB DL, RRC connected users, and DRB. The information can reflect the current status or represent forecast values.
[0028] For example, multi-carrier resource sharing configuration information may indicate the RAN's ability to set up dual connectivity or carrier aggregation with overlapping coverage of different frequencies available on the same slice. Slice radio resource management (RRM) policy or restriction information may indicate, for example, policies (e.g., by policy number) used for slice radio resource management or restrictions for slice radio resource management. Resource deployment information may indicate, for example, the deployment of slice resources by frequency. Slice-based cell reselection information may indicate, for example, the ability of a network node or device to support slice-based cell reselection. Slice service level agreement (SLA) information may indicate, for example, a service agreement between a provider and a user. PDU (Protocol Data Unit) session quality of service (QoS) information may include, for example, GBR (Guaranteed Bitrate), non-GBR, slice MBR (Maximum Bitrate), and / or service type. Information on attributes of used slice resources may indicate, for example, the preemption attribute of the slice resources used. Predicted service traffic information may indicate, for example, the predicted service of a network node. Dual connectivity configuration information indicates, for example, the dual connectivity capability or setup of a particular network node or device. Validity period information indicates, for example, the expiration or validity period of a predicted parameter, value, or configuration. Confidence information indicates, for example, the degree of accuracy or confidence of a predicted parameter, value, or configuration.
[0029] The input provided from a user equipment (UE) using slice resources to an AI / ML model trainer or AI / ML model inference unit (or intermediate data collector) (e.g., RAN slice-related measurement results) includes at least one or a combination of the following: UE location information, UE measurement information, or UE capability information. For example, UE location information may include at least one of the following: UE coordinates, UE serving cell ID, or UE moving velocity. UE measurement information may include at least one of the following: UE RSRP (reference signal received power) measurement, RSRQ (reference signal received quality) measurement, SINR (signal-to-interference noise ratio) measurement, or UE data throughput. UE capability information may include, for example, an indication of whether the UE supports a particular function. For example, this may indicate whether the UE supports slice-based cell reselection.
[0030] The inputs (e.g., support information) provided from one or more neighboring RANs (or neighboring (one or more) gNBs, (one or more) gNB-CUs, (one or more) gNB-DUs) to the AI / ML model trainer or AI / ML model inference unit (or intermediate data collector) include at least one or a combination of the following: second current or predicted slice availability information, second current or predicted shared network slice information, second current or predicted preferred network slice information, second current or predicted dedicated network slice information, second multi-carrier resource sharing configuration information, or second slice radio resource management (RRM) policy or restriction information. For example, current or predicted slice availability information indicates the capacity of available slice sources for a local network node. The information may, for example, reflect the current state or be a predicted value. Current or predicted shared network slice information may indicate the state of a shared network slice, such as the use of shared slices including PRBs, PRB ULs, PRB DLs, RRC connected users, and DRBs. The information can, for example, reflect the current state or represent predicted values. Current or predicted preferred network slice information can indicate the state of preferred network slices, such as the use of shared slices including PRB, PRB UL, PRB DL, RRC-connected users, and DRB. The information can, for example, reflect the current state or represent predicted values. The information can, for example, reflect the current state or represent predicted values. Current or predicted dedicated network slice information can indicate the state of preferred network slices, such as the use of shared slices including PRB, PRB UL, PRB DL, RRC-connected users, and DRB. The information can, for example, reflect the current state or represent predicted values.
[0031] For example, multi-carrier resource sharing configuration information may indicate the RAN's ability to set up dual connectivity or carrier aggregation with overlapping coverage of different frequencies available on the same slice. Slice radio resource management (RRM) policy or restriction information may indicate, for example, the policies (e.g., by policy number) used for slice radio resource management or restrictions for slice radio resource management.
[0032] One or more of the above information lists from the local RAN, neighboring RANs, or UEs can be used as inference feedback received from the actor in the model. The inference feedback information may include at least one or a combination of the following: shared resource usage feedback information, preferred resource usage feedback information, dedicated resource usage feedback information, service interruption feedback information, time-to-use feedback information, system performance feedback information, or (QoS) measurement feedback information. Exemplaryly, shared resource usage feedback information indicates the use of shared resources, preferred resource usage feedback information indicates, for example, the use of preferred resources, dedicated resource usage feedback information indicates, for example, the use of dedicated resources, service interruption feedback information indicates parameters related to service interruptions provided by network nodes, time-to-use feedback information indicates, for example, the time to use or expiration of feedback information, system performance feedback information may include one or more KPIs or other performance metrics of the actor, and QoS measurement feedback information indicates measurements of various QoS parameters.
[0033] Inference output In some examples, the inference output includes at least one or a combination of the following: for example, Random Access Network (RAN) slice allocation strategy output information indicating the RAN allocation strategy; for example, Cell handover or reselection strategy output information indicating the RAN handover or reselection strategy; for example, Predicted shared network slice output information indicating the predicted parameters or state of shared network slices; for example, Predicted preferred network slice output information indicating the predicted parameters or state of preferred network slices; for example, Predicted dedicated network slice output information indicating the predicted parameters or state of dedicated network slices; for example, Predicted remapping policy output information indicating the network resource remapping policy for network nodes; for example, Predicted service traffic output information indicating the predicted parameters or state of service traffic; for example, Expiry time output information indicating the validity period or expiration date of the above output information.
[0034] Examples of different arrangements The following disclosure describes the communication and operation between various network nodes to obtain information for further processes in order to train an AI / ML model, perform model inference, and generate a slice resource placement configuration / policy. Periodically, the training and inference network nodes may send requests to other network nodes, such as neighboring gNBs, for support information, including the information described above. In addition, one or more UEs may provide the UE information described above to the network node training the model or inference. When the model is trained or updated, the training network node may send update information to the inference network node, if the training node and inference node are different, to set up or update the AI / ML model on such node. When inference is performed, the inference node may send the output of the inference to an actor, such as a neighboring gNB. In response, the actor may send feedback to the training or inference node (via a data collector, if necessary) for future training of the model or inference.
[0035] Figure 3 shows a flowchart of wireless communication according to some embodiments of the present disclosure. In this example, model training and model inference are performed by a first gNB (gNB1), gNB1, where both the model trainer and the model inference unit are located in gNB1. The order of steps below and elsewhere in the present disclosure is illustrative, and certain steps may be rearranged, skipped, or performed iteratively or periodically.
[0036] Step 1: One or more UEs report RAN slice-related measurement results to gNB1. In this step, one or more UEs may report several measurements related to the use of RAN slice resources to gNB1 so that the measurement information can be used for model training or inference. According to some embodiments, the measurements include at least one of UE location information, UE measurement information, or UE capability information. Examples of this information are described above and are applicable herein unless otherwise described.
[0037] Step 2: gNB1 sends a request message to gNB2 (neighboring base station) to request information to assist in the model training performed by gNB1. The information in the request message may include at least one of the following items, namely an indication of the requested information that should be provided from the neighboring RAN (gNB2) to the first RAN of gNB1. For example, predefined messages XnAP HANDOVER REQUEST, XnAP AI / ML INFORMATION REQUEST, XnAP RESOURCE STATUS REQUEST, NGAP Uplink RAN Configuration Transfer, or NGAP Downlink RAN Configuration Transfer can be used to carry the request from gNB1 to gNB2. Alternatively or additionally, a new message may be introduced to request assistant information for the purpose of slice resource optimization.
[0038] Step 3: gNB2 transmits support information, for example, via XnAP (Xn Application Protocol) or NGAP (Next Generation Application Protocol) (via AMF), in response to a request. This allows gNB1 to perform model training. Support information from one or more other gNBs may include at least one of the following: second current or predicted slice available capacity information, second current or predicted shared network slice information, second current or predicted preferred network slice information, second current or predicted dedicated network slice information, second multi-carrier resource sharing configuration information, or second slice radio resource management (RRM) policy or restriction information. Examples of this information are discussed above and are applicable here unless otherwise described.
[0039] For example, predefined messages such as XnAP AI / ML INFORMATION RESPONSE, XnAP RESOURCE STATUS UPDATE, NGAP Uplink RAN Configuration Transfer, or NGAP Downlink RAN Configuration Transfer can be reused for this purpose. Alternatively or additionally, new messages may be introduced to request assistant information for the purpose of slice resource optimization.
[0040] Table 1 below shows an example of the IE (Information Element) design within a message provided from gNB2 to gNB1. [Table 1-1] [Table 1-2]
[0041] Step 4: gNB1 trains its model based on slice-related measurements received from the UE and at least one of the supporting information received from other neighboring gNBs in order to predict the slice configuration result. The training may generate or update the AI / ML model so that inference can generate the slice configuration of the RAN.
[0042] Step 5: The UE sends slice-related measurement results and supporting information to the gNB1, as in Step 1. In this step, the UE may update the slice-related measurement results to the gNB1. In some examples, updates to the slice-related measurement results can be provided to the gNB1 periodically.
[0043] Steps 6-7: In the case of model inference, gNB1 sends a request to gNB2 to seek support information for model inference. Details of the request and feedback are explained in Steps 2-3.
[0044] Step 8: gNB1 performs model inference based on slice-related measurements received from the UE and at least one of the supporting information received from other (one or more) neighboring gNBs to predict RAN slice resource allocation / configuration. The inference output may include configuration for slice resource allocation. For example, the inference output may include Random Access Network (RAN) slice allocation strategy output information, cell handover or reselection strategy output information, predicted shared network slice output information, predicted preferred network slice output information, predicted dedicated network slice output information, predicted remapping policy output information, predicted service traffic output information, or validity time output information. Examples of this information are discussed above and are applicable here unless otherwise described.
[0045] Step 9: gNB1 transmits the prediction information to one or more neighboring gNBs, such as gNB2, via XnAP or NGAP. In some examples, the inference output includes at least one or a combination of the following: Random Access Network (RAN) slice allocation strategy output information, cell handover or reselection strategy output information, prediction shared network slice output information, prediction preferred network slice output information, prediction dedicated network slice output information, prediction remapping policy output information, prediction service traffic output information, or validity time output information. Examples of this information are discussed above and are applicable here unless otherwise described.
[0046] Step 10: gNB2 sends inference feedback information to gNB1. The inference feedback may include at least one or a combination of the following: shared resource usage feedback information, preferred resource usage feedback information, dedicated resource usage feedback information, service interruption feedback information, time-of-use feedback information, service interruption time, system performance feedback information, or (QoS) measurement feedback information. Examples of this information are described above and are applicable here unless otherwise described.
[0047] Figure 4 shows another flowchart of wireless communication according to some embodiments of the present disclosure. In this example, model training is performed by OAM / CN (core network), and model inference is performed by a first gNB (gNB1).
[0048] Step 1: One or more UEs report RAN slice-related measurement results to the gNB-CU. In this step, one or more UEs may report several measurements related to the use of RAN slice resources to the gNB-CU so that the measurement information can be used for model training or inference. According to some embodiments, the measurements include at least one of UE location information, UE measurement information, or UE capability information. Examples of this information are described above and are applicable herein unless otherwise described.
[0049] Step 2: The OAM / CN sends a request message to gNB1 (neighboring base station) to request information to support the model training being conducted by the OAM. The information in the request message may include at least one of the following items, namely an indication of the requested information that should be provided from the neighboring RAN to the first RAN of gNB1.
[0050] Step 3: gNB1 sends support information to OAM / CN, for example, in response to a request. This allows OAM / CN to perform model training. Support information from one or more other gNBs may include at least one of the following items: first current or predicted slice available capacity information, first current or predicted shared network slice information, first current or predicted preferred network slice information, first current or predicted dedicated network slice information, first multi-carrier resource sharing configuration information, first slice radio resource management (RRM) policy or restriction information, resource deployment information, slice-based cell reselection information, slice service level agreement information, PDU session quality of service (QoS) information, information on the attributes of slice resources used, predicted service traffic information, dual connectivity configuration information, validity time information, or confidence information. Examples of this information are discussed above and are applicable here unless otherwise described. An example of IE (Information Element) design in a message provided from gNB1 to OAM / CN is shown in Table 1 above.
[0051] Step 4: The OAM / CN trains the model based on slice-related measurements received from the UE and at least one of the supporting information received from other neighboring gNBs in order to predict the slice configuration results. The training may generate or update the AI / ML model so that the inference generates the slice configuration of the RAN.
[0052] Step 5: OAM / CN sends the AI / ML model update or configuration to gNB1 for updating or setting up the AI / ML model on gNB1.
[0053] Step 5a: The UE sends slice-related measurement results and supporting information to the gNB1, as in Step 1. In this step, the UE may update the slice-related measurement results to the gNB1. In some examples, updates to the slice-related measurement results can be provided to the gNB1 periodically.
[0054] Step 6: gNB1 sends a request message to gNB2 (neighboring base station) to request information to assist in the model training performed by gNB1. The information in the request message may include at least one of the following items, namely an indication of the requested information that should be provided from the neighboring RAN (gNB2) to the first RAN of gNB1. For example, the messages XnAP HANDOVER REQUEST, XnAP AI / ML INFORMATION REQUEST, XnAP RESOURCE STATUS REQUEST, NGAP Uplink RAN Configuration Transfer, and NGAP Downlink RAN Configuration Transfer can be used to carry requests from gNB1 to gNB2. Alternatively or additionally, new messages may be introduced to request assistant information for the purpose of slice resource optimization.
[0055] Step 7: gNB2 transmits support information, for example, in response to a request, via XnAP (Xn Application Protocol) or NGAP (Next Generation Application Protocol) (via AMF). This allows gNB1 to perform model inference. Support information from one or more other gNBs may include at least one of the following: second current or predicted slice available capacity information, second current or predicted shared network slice information, second current or predicted preferred network slice information, second current or predicted dedicated network slice information, second multi-carrier resource sharing configuration information, or second slice radio resource management (RRM) policy or restriction information. Examples of this information are discussed above and are applicable here unless otherwise described.
[0056] As an example, predefined messages such as XnAP AI / ML INFORMATION RESPONSE, XnAP RESOURCE STATUS UPDATE, NGAP Uplink RAN Configuration Transfer, or NGAP Downlink RAN Configuration Transfer can be reused for that purpose. Alternatively or additionally, new messages may be introduced to request assistant information for the purpose of slice resource optimization. An example of IE (Information Element) design in a message provided from gNB2 to gNB1 is shown in Table 1 above.
[0057] Step 8: gNB1 performs model inference based on slice-related measurements received from the UE and at least one of the supporting information received from other (one or more) neighboring gNBs to predict RAN slice resource allocation / configuration. The inference output may include configuration for slice resource allocation. In some examples, the inference output may include at least one or a combination of the following: Random Access Network (RAN) slice allocation strategy output information, cell handover or reselection strategy output information, predicted shared network slice output information, predicted preferred network slice output information, predicted dedicated network slice output information, predicted remapping policy output information, predicted service traffic output information, or validity time output information.
[0058] Step 9: gNB1 transmits the prediction information to one or more neighboring gNBs, such as gNB2, via XnAP or NGAP. In some examples, the inference output includes at least one or a combination of the following: Random Access Network (RAN) slice allocation strategy output information, cell handover or reselection strategy output information, prediction shared network slice output information, prediction preferred network slice output information, prediction dedicated network slice output information, prediction remapping policy output information, prediction service traffic output information, or validity time output information. Examples of this information are discussed above and are applicable here unless otherwise described.
[0059] Step 10: gNB2 sends inference feedback information to gNB1. The inference feedback may include at least one or a combination of the following: shared resource usage feedback information, preferred resource usage feedback information, dedicated resource usage feedback information, service interruption feedback information, time-of-use feedback information, service interruption time, system performance feedback information, or (QoS) measurement feedback information. Examples of this information are described above and are applicable here unless otherwise described.
[0060] Figure 5 shows another flowchart of wireless communication according to some embodiments of the present disclosure. In this example, the gNB is configured as a gNB-CU and gNB-DU partitioned configuration. In this example, model training and model inference are performed by the gNB-CU, where both the model trainer and the model inference unit are located in the gNB-CU.
[0061] Step 1: One or more UEs report RAN slice-related measurement results to the gNB-CU. In this step, one or more UEs may report several measurements related to the use of RAN slice resources to the gNB-CU so that the measurement information can be used for model training or inference. According to some embodiments, the measurements include at least one of UE location information, UE measurement information, or UE capability information. Examples of this information are described above and are applicable herein unless otherwise described.
[0062] Step 2: The gNB-CU sends a request message to the gNB-DU to request information to assist in the model training performed by the gNB-CU. The information in the request message may include at least one of the following items, namely an indication of the requested information that the gNB-DU should provide to the gNB-CU. In one example, the message may be sent via F1AP (F1 Application Protocol). Alternatively or additionally, a new message may be introduced to request assistance information for the purpose of slice resource optimization.
[0063] Step 3: The gNB-DU transmits support information, for example, via F1AP in response to a request. This allows gNB1 to perform model training. Support information from one or more other gNB-DUs may include at least one of the following items: second current or predicted slice available capacity information, second current or predicted shared network slice information, second current or predicted preferred network slice information, second current or predicted dedicated network slice information, second multi-carrier resource sharing configuration information, or second slice radio resource management (RRM) policy or restriction information. Examples of IE (Information Element) designs in messages provided from a gNB-DU to a gNB-CU are shown in Table 1 above. Examples of this information are discussed above and are applicable here unless otherwise described.
[0064] Step 4: The gNB-CU trains its model based on slice-related measurements received from the UE and at least one of the supporting information received from other gNB-DUs in order to predict the slice configuration results. The training may generate or update the AI / ML model so that inference can generate the slice configuration of the RAN.
[0065] Step 5: The UE transmits slice-related measurement results and supporting information to the gNB-CU, similar to Step 1. In this step, the UE may update the slice-related measurement results to the gNB-CU. In some examples, updates to the slice-related measurement results can be provided to the gNB-CU periodically.
[0066] Steps 6-7: In the case of model inference, the gNB-CU sends a request to the gNB-DU to seek support information for model inference via F1AP. Details of the request and feedback are explained in Steps 2-3.
[0067] Step 8: The gNB-CU performs model inference based on slice-related measurements received from the UE and at least one of the supporting information received from other gNB-DUs to predict RAN slice resource allocation. The inference output may include a configuration for slice resource allocation.
[0068] Step 9: The gNB-CU transmits the prediction information to one or more gNB-DUs via the F1AP. In some examples, the inference output includes at least one or a combination of the following: Random Access Network (RAN) slice allocation strategy output information, cell handover or reselection strategy output information, prediction shared network slice output information, prediction preferred network slice output information, prediction dedicated network slice output information, prediction remapping policy output information, prediction service traffic output information, or validity period output information. Examples of this information are discussed above and are applicable here unless otherwise noted.
[0069] Step 10: The gNB-DU sends inference feedback information to the gNB-CU, for example, via the F1AP. The inference feedback may include at least one or a combination of the following: shared resource usage feedback information, preferred resource usage feedback information, dedicated resource usage feedback information, service interruption feedback information, time-of-use feedback information, service interruption time, system performance feedback information, or (QoS) measurement feedback information. Examples of this information are described above and are applicable here unless otherwise described.
[0070] Figure 6 shows another flowchart of wireless communication according to some embodiments of the present disclosure. In this example, the gNB is configured as a gNB-CU and gNB-DU partitioned arrangement. In this example, model training can be performed by the gNB-CU and model inference can be performed by the gNB-DU.
[0071] Step 1: One or more UEs report RAN slice-related measurement results to the gNB-CU. In this step, one or more UEs may report several measurements related to the use of RAN slice resources to the gNB-CU so that the measurement information can be used for model training or inference. According to some embodiments, the measurements include at least one of UE location information, UE measurement information, or UE capability information. Examples of this information are described above and are applicable herein unless otherwise described.
[0072] Step 2: The gNB-CU sends a request message to the gNB-DU to request information to assist in the model training performed by the gNB-CU. The information in the request message may include at least one of the following items, namely an indication of the requested information that the gNB-DU should provide to the gNB-CU. In one example, the message may be sent via F1AP (F1 Application Protocol). Alternatively or additionally, a new message may be introduced to request assistance information for the purpose of slice resource optimization.
[0073] Step 3: The gNB-DU transmits support information, for example, via F1AP in response to a request. This allows gNB1 to perform model training. Support information from one or more other gNB-DUs may include at least one of the following: second current or predicted slice available capacity information, second current or predicted shared network slice information, second current or predicted preferred network slice information, second current or predicted dedicated network slice information, second multi-carrier resource sharing configuration information, or second slice radio resource management (RRM) policy or restriction information. Examples of this information are discussed above and are applicable here unless otherwise described.
[0074] Table 1 above shows an example of the IE (Information Element) design within a message provided from gNB-DU to gNB-CU.
[0075] Step 4: The gNB-CU trains its model based on slice-related measurements received from the UE and at least one of the supporting information received from other gNB-DUs in order to predict the slice configuration results. The training may generate or update the AI / ML model so that inference can generate the slice configuration of the RAN.
[0076] Step 5: The UE transmits slice-related measurement results and supporting information to the gNB-CU, similar to Step 1. In this step, the UE may update the slice-related measurement results to the gNB-CU. In some examples, updates to the slice-related measurement results can be provided to the gNB-CU periodically.
[0077] Step 6: The gNB-CU transfers the received slice-related measurement results from the UE to the gNB-DU.
[0078] Step 7: The gNB-DU sends a request message to the gNB-CU to request information to assist the model inference performed by the gNB-DU. The information in the request message may include at least one of the following items, namely an indication of the requested information that the gNB-CU should provide to the gNB-DU. In one example, the message may be sent via F1AP (F1 Application Protocol). Alternatively or additionally, a new message may be introduced to request assistant information for the purpose of slice resource optimization.
[0079] Step 8: The gNB-CU sends support information, for example, via F1AP in response to a request. This allows the gNB-DU to perform model inference. The support information from the gNB-CU may include at least one of the following items: second current or predicted slice available capacity information, second current or predicted shared network slice information, second current or predicted preferred network slice information, second current or predicted dedicated network slice information, second multi-carrier resource sharing configuration information, or second slice radio resource management (RRM) policy or restriction information. Examples of this information are discussed above and are applicable here unless otherwise described. An example of IE (Information Element) design in a message provided from the gNB-DU to the gNB-CU is shown in Table 1 above.
[0080] Step 9: The gNB-DU performs model inference based on slice-related measurements received from the UE and at least one of the supporting information received from other gNB-CUs to predict RAN slice resource allocation / configuration. The inference output may include configuration for slice resource allocation.
[0081] Step 10: The gNB-DU sends the prediction information to the gNB-CU via the F1AP. In some examples, the inference output includes at least one or a combination of the following: Random Access Network (RAN) slice allocation strategy output information, cell handover or reselection strategy output information, prediction shared network slice output information, prediction preferred network slice output information, prediction dedicated network slice output information, prediction remapping policy output information, prediction service traffic output information, or validity period output information. Examples of this information are discussed above and are applicable here unless otherwise described.
[0082] Step 11: The gNB-CU sends inference feedback information to the gNB-DU, for example, via F1AP. The inference feedback may include at least one or a combination of the following: shared resource usage feedback information, preferred resource usage feedback information, dedicated resource usage feedback information, service interruption feedback information, time-of-use feedback information, service interruption time, system performance feedback information, or (QoS) measurement feedback information. Examples of this information are described above and are applicable here unless otherwise described.
[0083] The various configurations described above can be used for different purposes to best suit the application. Configuring the operation instructions or slice resource allocation using AI / ML models allows for efficient use of resources, taking into account the collected data and trained models.
[0084] According to some embodiments of this disclosure, a wireless communication method is disclosed. The method includes acquiring input information and using the input information to perform at least one of the following: model training of a slice wireless resource management model or generation of an inference output using a slice wireless resource management model. The input information includes at least one of slice measurement result information, one or more first pieces of information, one or more second pieces of information, or inference feedback information. The slice measurement result information includes at least one of UE location information, UE measurement information, or UE capability information. One or more first pieces of information include at least one of the following: first current or predicted slice availability information, first current or predicted shared network slice information, first current or predicted preferred network slice information, first current or predicted dedicated network slice information, first multi-carrier resource sharing configuration information, first slice radio resource management (RRM) policy or restriction information, resource deployment information, slice-based cell reselection information, slice service level agreement information, PDU session quality of service (QoS) information, information on attributes of slice resources used, predicted service traffic information, dual connectivity configuration information, validity time information, or reliability information. One or more second pieces of information include at least one of the following: second current or predicted slice availability information, second current or predicted shared network slice information, second current or predicted preferred network slice information, second current or predicted dedicated network slice information, second multi-carrier resource sharing configuration information, or second slice radio resource management (RRM) policy or restriction information.
[0085] According to some examples, the inference output includes at least one of the following: random access network (RAN) slice allocation strategy output information, cell handover or reselection strategy output information, predictive shared network slice output information, predictive preferred network slice output information, predictive dedicated network slice output information, predictive remapping policy output information, predictive service traffic output information, or validity period output information.
[0086] According to some examples, inference feedback information includes at least one of the following: shared resource usage feedback information, preferred resource usage feedback information, dedicated resource usage feedback information, service interruption feedback information, time-of-use feedback information, service interruption time, system performance feedback information, or (QoS) measurement feedback information.
[0087] According to some examples, performing at least one of model training or generating inference output using input information includes performing model training and generating inference output by a first network node. The method further includes receiving slice measurement results information from a user device (UE), transmitting the inference output to a second network node, and receiving inference feedback information from the second network node.
[0088] According to some examples, the method further includes sending a first request from a first network node to a second network node for first inference model input information for training, and sending a second request from the first network node to the second network node for first inference model input information for generating an inference output.
[0089] In some examples, the first network node is the first base station, and the second network node is the second base station.
[0090] According to some examples, the first network node is a centralized base station unit, and the second network node is a distributed base station unit.
[0091] According to some examples, performing at least one of model training or generating inference output using input information includes performing model training by a third network node.
[0092] According to some examples, the method further includes sending model deployment or update information to the first network node in order to prepare a slice radio resource management model that the first network node uses to generate inference output, in accordance with the results of training by the third network node.
[0093] According to some examples, the third network node houses the OAM (Operations, Management, and Maintenance) unit of the core network.
[0094] In some examples, performing at least one of model training or inference output generation using input information includes generating an inference output by a first network node based on a slice radio resource management model trained by a third network node.
[0095] According to some examples, the method further includes sending a first request from a first network node to a second network node for one or more first pieces of information to generate an inference output, and sending an inference output from the first network node to the second network node.
[0096] According to some examples, performing at least one of model training or generating inference output using input information includes training the model by a first network node. The method further includes the first network node transmitting model deployment or update information to a second network node in accordance with the results of the training by the first network node, in order to prepare a slice radio resource management model that the second network node will use to generate inference output.
[0097] In some examples, using input information to perform at least one of model training or inference output generation for a slice radio resource management model includes generating an inference output by a second network node based on a slice radio resource management model trained by a first network node.
[0098] According to some examples, the method further includes sending a first request from a second network node to a first network node for one or more first pieces of information to generate an inference output, and sending an inference output from the second network node to the first network node.
[0099] According to some examples, the first network node is a centralized base station unit, and the second network node is a distributed base station unit.
[0100] Figure 7 illustrates a block diagram of an exemplary wireless communication system 10 according to several embodiments of the present disclosure. System 10 may perform the methods / steps and combinations thereof disclosed herein. System 10 may include components and elements configured to support operational features that do not need to be described in detail herein.
[0101] System 10 may include a base station (BS) 110 and user equipment (UE) 120. BS 110 includes a BS transceiver or transceiver module 112, a BS antenna system 116, a BS memory or memory module 114, a BS processor or processor module 113, and a network interface 111. The components of BS 110 may be electrically coupled to each other and communicate as needed via a data communication bus 180. Similarly, UE 120 includes a UE transceiver or transceiver module 122, a UE antenna system 126, a UE memory or memory module 124, a UE processor or processor module 123, and an I / O interface 121. The components of UE 120 may be electrically coupled to each other and communicate as needed via a data communication bus 190. BS 110 communicates with UE 120 via a communication channel, which can be any wireless channel or other medium known in the art suitable for data transmission as described herein. The channel may contain the carriers of PCell and SCell.
[0102] Processor modules 113, 123 may be implemented or realized using general-purpose processors, associative memory, digital signal processors, application-specific integrated circuits, field-programmable gate arrays, any suitable programmable logic devices, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. Thus, processor modules may be realized as microprocessors, controllers, microcontrollers, state machines, etc. Processor modules may also be implemented as combinations of computing devices, for example, a combination of a digital signal processor and a microprocessor, multiple microprocessors, one or more microprocessors working in conjunction with a digital signal processor core, or any other such configuration.
[0103] Furthermore, the steps of the methods or algorithms described in relation to the embodiments disclosed herein may be embodied, respectively, directly in hardware, in firmware, in software modules performed by processor modules 113, 123, or in any practical combination thereof. Memory modules 113, 123 may be implemented as RAM memory, flash memory, EEPROM memory, registers, ROM memory, EPROM memory, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art. In this regard, memory modules 114, 124 may be coupled to processor modules 113, 123, respectively, so that processor modules 113, 123 can read information from and write information to memory modules 114, 124, respectively. Memory modules 114, 124 may also be incorporated into their respective processor modules 113, 123. In some embodiments, memory modules 114, 124 may each include cache memory for storing temporary variables or other intermediate information during the execution of instructions to be performed by processor modules 113, 123, respectively. Memory modules 114 and 124 may also each include non-volatile memory for storing instructions to be performed by processor modules 113 and 123, respectively.
[0104] Various exemplary embodiments of the Disclosure are described herein with reference to accompanying drawings to enable those skilled in the art to create and use the Disclosure. The Disclosure is not limited to the exemplary embodiments and uses described and illustrated herein. Furthermore, the particular order and / or hierarchy of steps in the methods disclosed herein is merely an exemplary approach. Based on design preferences, the particular order or hierarchy of steps in the methods or processes disclosed may be rearranged while remaining within the scope of the Disclosure. Thus, those skilled in the art will understand that the methods and techniques disclosed herein present various steps or operations in an exemplary (one or more) order, and that the Disclosure is not limited to the particular order or hierarchy presented unless otherwise specified.
[0105] This disclosure is intended to cover any conceivable variations, uses, combinations, or adaptive modifications of this disclosure in accordance with the general principles of this disclosure, and includes well-known knowledge and prior art means in the art that are not disclosed in this application.
[0106] This disclosure is not limited to the exact structure or operation described above and shown in the accompanying drawings, and it should be understood that various modifications and changes may be made without departing from the scope of this application. The scope of this application is limited to the claims attached.
[0107] The methods, devices, processes, circuits, and logic described above can be implemented in many different ways and as many different combinations of hardware and software. For example, all or part of the implementation may be a circuit including an instruction processor or controller such as a central processing unit (CPU), a microcontroller, or a microprocessor; or a circuit as an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a field-programmable gate array (FPGA); or a circuit including discrete logic or other circuit components, including analog circuit components, digital circuit components, or both; or any combination thereof. The circuit may include, for example, individual interconnected hardware components, be combined on a single integrated circuit die, distributed across multiple integrated circuit dies, or implemented in a multi-chip module (MCM) of multiple integrated circuit dies in a common package.
[0108] Therefore, a circuit may store or access instructions for execution, or its functionality may be implemented solely in hardware. Instructions may be stored in tangible storage media other than temporary signals, such as flash memory, random-access memory (RAM), read-only memory (ROM), or erasable programmable read-only memory (EPROM), or on magnetic or optical disks, such as compact disk read-only memory (CDROM), hard disk drives (HDDs), or on other magnetic or optical disks, or on other machine-readable media. Products such as computer program products may include a storage medium and instructions stored in or on the medium, and when performed by a circuit within the device, the instructions may cause the device to implement any of the processes described above or illustrated in the drawings.
[0109] Implementations can be distributed. For example, a circuit may include multiple separate system components such as multiple processors and memories, or it may extend to multiple distributed processing systems. Parameters, databases, and other data structures may be stored and managed separately, incorporated into a single memory or database, logically and physically organized in many different ways, and implemented in many different ways. Exemplary implementations include linked lists, program variables, hash tables, arrays, records (e.g., database records), objects, and implicit storage mechanisms. Instructions may form part of a single program (e.g., a subroutine or other code section), form multiple separate programs, be distributed across multiple memories and processors, and be implemented in many different ways. Exemplary implementations include standalone programs and those included as part of a library, such as a shared library like a dynamic-link library (DLL). A library may include, for example, shared data and one or more shared programs that, when performed by the circuit, perform any of the operations described above or illustrated in the diagrams.
[0110] In some examples, each unit, subunit, and / or module of a system may contain a logical component. Each logical component may be hardware or a combination of hardware and software. For example, each logical component may include an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), a combination of digital logic circuits, analog circuits, discrete circuits, gates, or any other type of hardware or combination thereof. Alternatively or additionally, each logical component may include memory hardware, such as a portion of memory containing instructions executable by a processor or other processor to implement one or more of the logical component's features. If any one of the logical components contains a portion of memory containing instructions executable by a processor, the logical component may or may not include a processor. In some examples, each logical component may be a portion of memory or other physical memory containing instructions executable by a processor or other processor to implement the corresponding logical component's features without the logical component including other hardware. Even if the included hardware includes software, each logical component includes at least some hardware, so each logical component may be called interchangeably with a hardware logical component.
[0111] A second action can be said to be "in response" to the first action, regardless of whether the second action arises directly or indirectly from the first action. The second action may occur substantially later than the first action and may also be in response to the first action. Similarly, a second action can be said to be in response to the first action even if there are intervening actions between the first and second actions, and one or more of these intervening actions directly cause the second action to occur. For example, if the first action sets a flag, and each time the flag is set, a third action subsequently initiates the second action, then the second action may be in response to the first action.
[0112] To clarify its use and thereby inform the public, 、 , ...and <n> at least one of the following" or< / n> 、 、… <n> or at least one of those combinations" or "< / n> 、 , ...and / or <n> The phrase "..." is defined by the Applicant in its broadest sense and supersedes any other implicit definitions before or after it to mean one or more elements selected from the group including A, B, ..., and N, unless expressly asserted otherwise by the Applicant. In other words, these phrases mean one element in any combination of one or more elements A, B, ..., or N, which may contain only one element, or one or more other elements which may also contain additional elements not listed.< / n>
Claims
1. A wireless communication method, The process involves obtaining input information, wherein the input information is Slice measurement result information, UE location information, UE measurement information, or UE capability information Slice measurement result information including at least one of the following, One or more first pieces of information, First, current or predicted slice available capacity information, The first current or predicted shared network slice information, First, current or predicted priority network slice information, The first current or forecast-only network slice information, First multi-carrier resource sharing configuration information, The first slice of radio resource management (RRM) policy or restriction information, Resource deployment information, Slice-based cell reselection information, Slice service level agreement information, PDU session service quality (QoS) information, Information on the attributes of the sliced resources used, Predictive service traffic information, Dual connectivity configuration information, Validity period information, or Reliability information One or more pieces of first information, including at least one of the following: One or more second pieces of information, Second, current or projected slice availability information, Second current or predicted shared network slice information, Second current or predicted priority network slice information, Second, current or forecast-only network slice information, Second multi-carrier resource sharing configuration information, or Second slice Radio Resource Management (RRM) policy or restriction information, One or more pieces of second information, including at least one of the following, or Inference feedback information It includes at least one of the following, Using the aforementioned input information, perform at least one of the following: model training of a slice wireless resource management model or generation of inference output using the slice wireless resource management model. Methods that include...
2. The aforementioned inference output is, Random Access Network (RAN) slice allocation strategy output information. Cell handover or reselection strategy output information, Predictive shared network slice output information, Prediction-prioritized network slice output information, Prediction-only network slice output information, Predictive remapping policy output information, Predictive service traffic output information, or Effective time output information The method according to claim 1, comprising at least one of the following.
3. The aforementioned inference feedback information is Shared resource usage feedback information, Priority resource usage feedback information, Dedicated resource usage feedback information, Service interruption feedback information, Effective time feedback information, Service interruption time, System performance feedback information, or QoS measurement value feedback information The method according to claim 1, comprising at least one of the following.
4. Performing at least one of model training or generating the inference output using the input information includes performing the model training and generating the inference output by a first network node, and the method is Receiving the slice measurement result information from the user equipment (UE), The output of the inference is transmitted to a second network node, Receiving the inference feedback information from the second network node and The method according to claim 1, further comprising:
5. The first network node transmits a first request to the second network node for the one or more second pieces of information for the training, The first network node transmits a second request to the second network node for one or more second pieces of information to generate the inference output. The method according to claim 4, further comprising:
6. The method according to claim 4 or 5, wherein the first network node is a first base station, and the second network node is a second base station.
7. The method according to claim 4 or 5, wherein the first network node is a base station centralization unit and the second network node is a base station distribution unit.
8. The method according to claim 1, wherein performing at least one of training a model or generating the inference output using the input information includes performing the model training by a third network node.
9. The method according to claim 8, further comprising transmitting model deployment or update information to the first network node in order to prepare the slice radio resource management model that the first network node uses to generate the inference output, in accordance with the results of the training by the third network node.
10. The method according to claim 8, wherein the third network node comprises an OAM (operation, management, and maintenance) unit of the core network.
11. The method according to claim 1, wherein performing at least one of model training or generating the inference output using the input information includes generating the inference output by the first network node based on the slice radio resource management model trained by the third network node.
12. The first network node transmits a first request to the second network node for one or more first pieces of information to generate the inference output, The first network node transmits the inference output to the second network node. The method according to claim 11, further comprising:
13. Performing at least one of model training or generating the inference output using the input information includes performing the model training by a first network node, and the method is In accordance with the training results by the first network node, the first network node transmits model deployment or update information to the second network node in order to prepare the slice radio resource management model that the second network node uses to generate the inference output. The method according to claim 1, further comprising:
14. The method according to claim 1, wherein using the input information to perform at least one of model training of a slice radio resource management model or generation of the inference output includes generating the inference output by a second network node based on the slice radio resource management model trained by a first network node.
15. The second network node transmits a first request to the first network node for one or more first pieces of information to generate the inference output, The second network node transmits the inference output to the first network node. The method according to claim 14, further comprising:
16. The method according to claim 14 or 15, wherein the first network node is a base station centralization unit and the second network node is a base station distribution unit.
17. A wireless communication device comprising one or more memory units for storing one or more programs, and one or more processors electrically coupled to the one or more memory units and configured to execute the one or more programs and perform any one of the methods according to claims 1 to 16 or a combination thereof.
18. A non-temporary computer-readable storage medium for storing one or more programs, wherein the one or more programs, when executed by at least one processor, are configured to perform any one of the methods according to claims 1 to 16 or a combination thereof.